### Table 1: Evaluating solutions on various levels of concretization by combining objects and methods

"... In PAGE 2: ... Only through the combination of objects and methods the problem solving process did occur. Table1 gives an example of the evaluation processes that occurred on various levels of concretization. During the tests, with growing experience, the test persons applied more concrete methods onto... ..."

### Table 7: Comparison of the average solution time per problem and the number of solved

1995

"... In PAGE 46: ....3.2 Results Wehave determined the solution time for each of the 100 problems in each of the described modes. The average solution time as well as the number of problems that could be solved within the time limit is shown in Table7 . Wehave determined these values for reasoning from abstract cases separately for each of the three types of abstract cases.... In PAGE 48: ... We can see that the best training sets leads to a problem solving time which is similar or only slightly worse than the optimum shown in Table 7. Even in the average case, considerable improvements over the pure search and hierarchical problem solving #28compare Table7 and Table 9#29 can be discovered. The same... In PAGE 51: ... It turns out that for all training sets, learning improves the concrete level problem solver, but that the speedup is much smaller than when using the original abstract problem solving domain #28cf. Table7 and 9#29. In particular, none of the resulting speedups over concrete level problem solving were signi#0Ccant.... In PAGE 51: ... There is still a slight improvement in the number of problems that could be solved after learning but the improvementismuch smaller than when using the original abstract problem solving domain #28cf. Table7 and 10#29. 17.... In PAGE 53: ... The abstract solutions determined by this procedure are useful for solving new concrete problems, because they have a high chance of being re#0Cnable. The detailed experimental evaluation with the fully implemented Paris system in the domain of mechanical engineering strongly demonstrates that Paris can signi#0Ccantly im- prove problem solving in situations in which a hierarchical problem solver using dropping sentences fails to showanadvantage #28see Table7 to 11#29. 10.... In PAGE 57: ... Furthermore, we will address the development of highly e#0Ecient retrieval algorithms for abstract cases. As Table7 shows, the retrieval mechanism has a strong in#0Duence on the achieved speedup. Even if the linear retrieval wehave presented turned out to be pretty good, we expect a utility problem #28Minton, 1990#29 to occur when the size of the case- base grows.... ..."

Cited by 3

### Table 1 Final Global Prediction Accuracies

in Key Words

"... In PAGE 18: ... Thus the space of possibilities is huge. For global predictions the actual accuracy ratings are small, and although the absolute change of performance is roughly the same for both concrete and abstract global prediction, the relative difference is far greater at the concrete level (see Table1 , below). Figs.... ..."

### Table 3: First vehicle then crew scheduling

1999

Cited by 1

### Table 3: First vehicle then crew scheduling

1999

Cited by 1

### Table 4 Classification results with different reducts 1: Number of rules; 2: Classification accuracy POSAR CEAR DISMAR GAAR PSORSAR

"... In PAGE 25: ... So, all the particles have a powerful search capability, which can help the swarm avoid dead ends. The comparison of the number of decision rules and the classification accuracy with different reducts are shown in Table4... ..."

### Table 1. Usability Activities by Source.

2003

"... In PAGE 4: ... The sources vary as to the extent of formalization. The set of usability-related activities proposed in the HCI field are detailed in Table1 , where sources follow an order of increasing formalization from left to right. We have analyzed the activities proposed by the different authors in order to extract the common ones or, at least, the activities that are at the same abstraction level and are common to several sources.... In PAGE 5: ...The resulting usability activities (the left column in Table1 ) are represented in Fig. 1, grouped according to the generic kind of activity to which they belong: Analy- sis, design or evaluation.... ..."

Cited by 1

### Table 5: Computational results for Example 1 for constant processing times. MILP STN Model MILP/CP Hybrid Scheme

2004

"... In PAGE 28: ...ith 13 time points, while instances Ms2 and Ms3 were solved for various time grids (i.e. no of time points) but no integer feasible solution was found in 36,000 CPU seconds. The computational statistics reported in Table5 , correspond to the MILP with the minimum number of time points that can represent the optimal solution found by the hybrid algorithm. The proposed algorithm, on the other hand, obtained the optimal solution for all instances of both objectives.... ..."

Cited by 7

### Table 2: Results for 30 di erent test problems.

"... In PAGE 15: ... i = 0; Stop := False; repeatfold B := fB, fold I := fI; do x 2 Neighbors(xi) evaluate x ! f(x); if f(x) gt; fB and x is feasible thenfB := f(x), xB := x; elseif f(x) gt; max(fB; fI) thenfI := f(x), xI := x; enddo if fB gt; fold B then xi+1 = xB; elseif fI gt; max(fold B ; fold I ) then xi+1 = xI; else Stop := True; i := i + 1; until Stop = True; if fB gt; 0 then Feasible Local Optimum found else No feasible solution found end.From the 30 test cases that are listed in Table2 , the restart from the best infeasible neighbor turned out to be helpful in almost half of the cases, as is shown in Table 1. On the other hand it shows that in a few cases it caused a longer search without improving the objective function.... In PAGE 16: ...04667 54 1.04670 62 Table 1: Test problems from Table2 where infeasible neighbors were explored. lowed is then gradually decreased, until only feasible patterns can be selected as new parents.... In PAGE 19: ... PI: Pairwise interchange from an arbitrary starting point. The results are listed in Table2 . Computation times are in seconds on an HP 9000/720 workstation.... ..."

### Table 3 Results for 30 di erent test problems.

"... In PAGE 19: ...04667 54 1.04670 62 Table 2 Test problems from Table3 where infeasible neighbors were explored. Of course, there are several other options possible to take advantage of infea- sible patterns.... In PAGE 21: ...2 Computational results We compared 4 algorithms: (1) DC: DICOPT running under the modeling language GAMS; (2) CR: CONOPT, followed by the simple rounding procedure; (3) CRP: As CR, but then followed by Pairwise Interchange; (4) PI: Pairwise interchange from an arbitrary starting point. The results are listed in Table3 . Computation times are in seconds on an HP 9000/720 workstation.... ..."